Software Engineer
Egor
Achkasov
MSc Computer Science, TU Dresden. Research background in systems programming, signal processing, and machine learning — with full-stack production delivery.
Career
Experience
Early team member at a pre-launch urban-tech startup applying computer vision to NYC’s public camera network for real-time street parking detection. Owned the complete technical stack: REST API, YOLOv8-based CV pipeline, Next.js admin dashboard, and all DevOps infrastructure. Microservices architecture orchestrated with Docker Compose; metrics via Prometheus.
Contributed to an IEEE 2025 paper on spectrum anomaly (jammer) detection via digital-twin channels. Calibrated electromagnetic material parameters in a Sionna ray-tracing scene using differentiable ray tracing and gradient-based optimisation (Adam), fitting simulated channel impulse responses to real-world wireless measurements from 3 access points across 512 subcarrier frequencies.
Full-stack contributions to HermesPy, an open-source Python/C++ radio system simulator used in
wireless research. Implemented DSP modules (signal access, MIMO channel fixes, array indexing),
a 3D ray-tracer in C, a ScaPy-based network frame generator, and CI/CD pipelines via GitLab.
Maintained 99% test coverage using the unittest framework throughout.
Built a 3D graphics engine and interactive visualisation tool for neuroscientists working with high-resolution brain scan data. Designed the full stack: OpenGL rendering engine, Qt/MVC GUI, and real-time data models for dynamic manipulation of high-polygon meshes. The work formed the basis of a Bachelor’s thesis benchmarking seven polygon mesh data structures (DCEL, BMesh, etc.) on construction, Catmull-Clark subdivision, traversal, and editing tasks.
Technical
Skills
Academic
Education
Open Source
Selected Projects
MIMO LMMSE detector targeting the RISC-V Vector Extension. Full pipeline: Gram matrix, Cholesky decomposition, forward/backward substitution — float and fixed-point.
Minimalistic signal-processing ray-tracer in C for simulating wireless channel propagation — geometry, ray-scene intersections, reflections, and CIR computation.
Material parameter calibration via differentiable ray tracing — associated with IEEE 2025 paper on spectrum jammer detection using digital twin channels.
TensorFlow- and Mitsuba-free reimplementation of NVIDIA Sionna’s ray-tracing pipeline. Removes heavyweight dependencies while preserving the simulation functionality.
Simulator for Communicating Finite-State Machine Systems — a formal model for distributed protocol verification and concurrent state-space exploration.
Manga page descrambler in Rust. Reverse-engineers and inverts tile-based image obfuscation from the COMIC Meteor platform — low-level binary image manipulation.
Credentials